Beyond the chart library
An interview with Jorge Camoes on making good dataviz, regardless of the tool
Does the tool we use to make charts matter? Do we need the priciest, shiniest software to communicate our data insights effectively? Or can Excel do the trick?
The above are questions a lot of people in the field struggle with. Some dataviz applications are considered superior, while others—like the poor old Excel—not fancy enough. So do you need a bright shiny tool to make good dataviz? To explore this question, I interviewed Jorge Camoes, known in the dataviz community for doing magic with tools that may have a bad rep.
Let’s dive into his wisdom.
Hi! Thank you so much for agreeing to talk to the readers of The Plot. Could you please briefly introduce yourself?
I am a consultant and a trainer mostly in the data visualization space for office users. That often includes working with Excel, PowerBI, or Tableau business data. I wrote a book, “Data at Work” which targets this specific group of professionals. I live in Lisbon, Portugal.
I love your tagline “dataviz for mortals”. How did it come about?
I have an unrequited love for graphic design. My drawing skills are legendary (in a bad sense) among friends and family. There are people though who can draw incredibly well. I marvel at the work of Adolfo Arranz, for example. His illustrations/visualisations are so stunning and informative! No one can convince me that there is no divine intervention there.
Fellow mortals and I can’t aspire to a fraction of that, but we should at least create a safety net to avoid basic design and data visualization faux pas. That was my goal when writing the book, during my training courses, and when working with clients.
You’re known in the dataviz community as one of the practitioners who can do miracles with Excel. What circumstances made you specialise in this tool and even write a book about it?
The word “Excel” is indeed in the book’s subtitle, but in hindsight, I think that is unfortunate and sends the wrong message. I wanted to write a book for Excel users because I thought they needed to go beyond the chart library and, as I said above, have a safety net that could protect them from falling into the many design traps in Excel. I made a point of making all the charts in Excel without any Adobe Illustrator edits. However, the tool is rarely mentioned in the book as I resisted writing a manual for Excel users.
As you probably know, the corporate culture is very Microsoft-friendly. I tried to work with other tools (R, for example) but couldn’t make them useful beyond a hobby. And Excel is a great tool for exploring visual ideas. You can create almost anything with a scatterplot, for example. But you probably need a different tool if you want something perfectly designed, something that you can easily interact with, or something that doesn’t take three hours to change instead of two seconds.
When making charts in Excel, assume there will be a lot of work replacing junk with useful visual elements. You can use what I call “design data” (data used to add line references, bands, highlight data points, etc).
What’s your reply to people who say that one can’t make good charts in Excel?
Give me an example of a good chart and there is a serious chance that an advanced Excel user can replicate it. It’s more about perception than actual limitations. And there is also a social dimension: do you want to disclose that that brilliant chart was made in… ugh… Excel?
You can’t hide the fact that most Excel users don’t have the right skills to make good charts and they/we accept defaults with abandon. If you use Tableau, R, or D3 you get the benefit of better defaults, and their users are often more aware of data visualisation as a field.
It’s true though that some features are missing or are hard to implement in Excel, like color encoding or small multiples. However, if you believe you can’t make good charts in Excel, you either don’t know the tool or you belong to a dataviz community with a different set of requirements.
What are your top tips for those among us who have to design charts with Excel and want to make them impactful?
As a rule of thumb, a chart should not be easily associated with a tool, and this applies to Excel but also to PowerBI, Tableau, or R. Changing defaults, like fonts and palettes, is the bare minimum. The Excel chart library is poor, so, when making charts in it, assume there will be a lot of work replacing junk with useful visual elements. You can use what I call “design data” (data used to add line references, bands, highlight data points, etc). Error bars are great for mimicking options that are not available.
Think of it this way: your starting chart covers 20% of what you’d like to design. Taking advantage of Excel’s flexibility will take you much further. But in the end, what makes a chart impactful has little to do with the software. If you don’t have clear ideas about the data, how to turn it into a message, and how that message is received and interpreted by the audience, no tool can do it for you. It can only make the process easier and more enjoyable.
What makes a chart impactful has little to do with the tool. If you don’t have clear ideas about the data, how to turn it into a message, and how it is received and interpreted by the audience, no tool can do it for you.
What’s the one thing your clients are most often surprised about when designing with Excel?
I could give you many examples related to out-of-the-box charts in Excel, but helping clients think visually and with intent is more rewarding to talk about. In a recent project, a client was using a probe with multiple sensors to measure something. They had a line chart where each series was a reading for a sensor, showing change over time. There’s nothing wrong with a line chart, of course. The problem is that since the sensors record a flow, where the sensor is located along the probe also matters. So they were missing a relevant pattern because it’s impossible to generate a mental image of the sequence and the patterns that emerge from it.
I suggested a different approach where they could see both the change over time and the pattern. Unfortunately, I can't share more details about this, but it's a similar process I used with these population pyramids years ago: how can I visualise a complex pattern while retaining the structure a demographer is used to? This is a common challenge as people tend to focus on which chart type to use, instead of what they need to see. And they’re often surprised by a change in perspective.
Finally, what advice do you have for our readers who are just starting? What do you wish you’d known at the very beginning of your data visualization journey?
Realising that there are no chart types, only a blank canvas where you can plot whatever you want, mix geometries, and encode data however you feel appropriate will be liberating. Then, you’ll need to step back and see if your approach makes sense, is appropriate for the specific situation, doesn’t break basic rules, etc.
Data visualisation communities are often defined by how they interpret form and function. You are not limited to a single interpretation, but it’s nice to find a place and a sense of belonging.
Additionally, respect but also torture the data at will, but don’t ignore it or use it as a prop. And statistics are fun, often more than the raw data.
And more important than everything else, never say “I hate pie charts”.
What stuck with you the most from this interview? For me, it’s the idea of not getting married to a given tool. We need to also develop other skills—hard and soft—to make great data visualisations, regardless of the software.
Thanks for reading The Plot! You can connect with Jorge on Twitter or LinkedIn.
And I will see you next week.
—Evelina
Want to develop your data storytelling skills?
Join my three-week online data storytelling bootcamp starting in early June. This will be an updated and improved version of my usual course. Use the code THEPLOT10 for 10% off.